There is a growing need for a theory of “local to global” in distributed multi-agent systems, one which is able systematically to describe and analyze a variety of problems. This is the first in a series of two papers that begins to develop such a theory. Here, we analyze one particular multi-agent problem – the “equigrouping problem,” in which multiple identical agents organize themselves into groups of equal size. We develop a formal model for describing the system and an notion of equivalence characterizing multi-agent algorithms in terms of the group behaviors induced by the algorithm. Our main result is a characterization of the space of all solutions to the equigrouping problem with respect to this group behavior equivalence. The result allows us to obtain infinitely many substantially different solutions to the Equigrouping problem, and to understand these different solutions in a qualitatively satisfying manner. The second paper in this series indicates how to de...